Mercurial > genshi > mirror
comparison genshi/tests/input.py @ 965:2bfd8f8d241c trunk
Fix parsing of multi-byte characters that occur on 4K boundaries of HTML files (fixes #538).
| author | hodgestar |
|---|---|
| date | Sat, 29 Dec 2012 13:02:20 +0000 |
| parents | 18209925c54e |
| children | 0f4b2e892a48 |
comparison
equal
deleted
inserted
replaced
| 963:99d4c481e4eb | 965:2bfd8f8d241c |
|---|---|
| 251 self.assertEqual(3, len(events)) | 251 self.assertEqual(3, len(events)) |
| 252 self.assertEqual((Stream.START, ('span', ())), events[0][:2]) | 252 self.assertEqual((Stream.START, ('span', ())), events[0][:2]) |
| 253 self.assertEqual((Stream.TEXT, "'"), events[1][:2]) | 253 self.assertEqual((Stream.TEXT, "'"), events[1][:2]) |
| 254 self.assertEqual((Stream.END, 'span'), events[2][:2]) | 254 self.assertEqual((Stream.END, 'span'), events[2][:2]) |
| 255 | 255 |
| 256 def test_multibyte_character_on_chunk_boundary(self): | |
| 257 text = u'a' * ((4 * 1024) - 1) + u'\xe6' | |
| 258 events = list(HTMLParser(BytesIO(text.encode('utf-8')), | |
| 259 encoding='utf-8')) | |
| 260 self.assertEqual(1, len(events)) | |
| 261 self.assertEqual((Stream.TEXT, text), events[0][:2]) | |
| 262 | |
| 256 | 263 |
| 257 def suite(): | 264 def suite(): |
| 258 suite = unittest.TestSuite() | 265 suite = unittest.TestSuite() |
| 259 suite.addTest(doctest.DocTestSuite(XMLParser.__module__)) | 266 suite.addTest(doctest.DocTestSuite(XMLParser.__module__)) |
| 260 suite.addTest(unittest.makeSuite(XMLParserTestCase, 'test')) | 267 suite.addTest(unittest.makeSuite(XMLParserTestCase, 'test')) |